Cytomegalovirus (CMV)-based vectors are uniquely capable of inducing long-term effector memory T cell responses and escaping vector-specific immunity. However, fully replicative vectors are unlikely to be approved for human use given the pathogenic potential of CMV, particularly in pregnant women and immunocompromised individuals. A central goal is therefore to increase vector safety while maintaining immunogenicity and efficacy. In this project, we will define how acute and persistent replication correlates with the ability of rhesus CMV to induce a T cell response to simian immunodeficiency virus (SIV). Preliminary data suggest that severely attenuated RhCMV can still induce CMV-specific immune responses in sero-negative animals consistent with persistent antigen production. Unlike the parental vector however, the such low-cycle vectors do not seem to be secreted from inoculated animals. Importantly, low-cycle vectors expressing simian immunedeficiency virus (SIV) antigens super-infect sero-positive animals and induce an SIV-specific T cell response suggesting retention of immunogenicity. In three specific aims we will determine the extent to which vectors can be rendered replication-incompetent without sacrificing immunogenicity and the pathogenicity of such vectors in the immunocompromised. Specific Aim 1 is to further reduce the ability of RhCMV/SIV vectors to replicate by constructing single-cycle vectors. In addition, we will construct vectors whose replication can be externally controlled thus allowing us to study the role of viral replication and spreading for the establishment and maintenance of persistent antigen production and T cell stimulation. Specific Aim 2 is to determine the pathogenicity of replication-deficient vectors in an animal model of congenital infection. A further aim is to compare their ability to induce robust SIV-specific effector memory T cell responses in sero-positive animals. Specific Aim 3 is to generate vectors combining the lowest pathogenicity with the highest immunogenicity for efficacy studies and to confirm that corresponding human CMV vectors are attenuated in a novel animal model of latent infection. Thus, we expect that the attenuation strategy developed here will be directly translatable into HCMV/HIV vectors.